YOLOv8 Based on Data Augmentation for MRI Brain Tumor Detection

Abstract

Purpose: This research aimed to detect meningioma, glioma, and pituitary brain tumors using the YOLOv8 architecture and data augmentations.

Methods: This research employed the YOLOv8 architecture with data augmentation techniques to detect meningioma, glioma, and pituitary brain tumors. The study collected a dataset of T1-weighted contrast-enhanced images. The dataset is used for training, validation, and testing. Preprocessing and augmentation are applied to enhance the training data.

Result: After applying data augmentation techniques, the performance of all tumor types improves significantly. Meningioma, Glioma, and Pituitary tumors demonstrate increased Precision, Recall, and mAP50 scores compared to the results before augmentation. The findings highlight the effectiveness of the proposed method in enhancing the model's ability to accurately detect brain tumors in MRI scans. The research conducted both with and without augmentation followed a similar procedure: data collection was first undertaken, followed by preprocessing and with or without augmentation. Subsequently, the collected data was partitioned into training and validation subsets for training with the YOLOv8 architecture. Finally, the model's performance was evaluated through testing to assess its effectiveness in detecting brain tumors.

Novelty: The novelty of this research lies in the YOLOv8 architecture and data augmentation techniques for MRI brain tumor detection. The study contributes to the existing knowledge by demonstrating the effectiveness of deep learning-based approaches in automating the detection process and improving the model's performance. By combining YOLOv8 with data augmentation, the proposed method enhances the model's accuracy and efficiency. The research findings emphasize the potential of this approach in facilitating early diagnosis and treatment planning, thereby improving patient care in the context of brain tumor detection.